Data Collection Strategies
Selecting a storm
The OC will choose storms based largely on coordination of data collection among all projects ongoing in the Hazardous Weather Testbed. During many events this will require an open phone line between the developmental lab and the HWT room. In general, storms will be sampled based on the following order of priority:
If no other HWT activities are collecting data, SHAVE is free to choose any storms of interest over the US. The OC should target storm types that have not been significantly sampled during the project. We desire to collect a diverse dataset of storm types in varying environmental conditions. However, in general, the severe weather threat type should be used as guidance for what type of storms and what areas to call on:
Primarily, wind and flash flooding calls can/will occur a day to several days after the event. This is primarily because of evacuations and any damage to telephone infrastructure. "Real-time" wind reports should only be taken if details are available. In other words, reports of hail with "strong winds" are just hail reports and not hail and wind reports. For wind and flash flooding surveys, local storm reports (LSRs) and flash flood guidance should be used to direct the survey.
We will sample both warned and unwarned storms. Data will be made available to those NWS offices that desire to examine the data in real-time. We would encourage those NWS offices that do use our data feed, however, to continue to collect information from their own storm spotter networks as per their normal operations. There is no guarantee that SHAVE will sample storms in a given NWS County Warning Area on a particular day, and even if we do so it is possible that we will only do so for a short time or may miss something that a regular spotter might report.
Collecting data on a stormIn general, phone calls in the immediate aftermath of a storm will focus on hail data collection, but participants will also ask about any high winds, wind damage, flash flooding or tornado sightings. If a significant swath of wind and/or tornado damage might have occurred with a storm of interest, follow up data collection using phone surveys and media data mining might be needed.
The general idea in collecting reports can be described as a "pocket" technique. While the ideal situation is to collect reports at an optimal spacing of 1 km. However, due rural areas or disconnect numbers, this may not be possible. Instead, reports should be collected in pockets with a starting radius of 1 km and moving outward. This will help ensure that an area is at least covered by 1 report and limit the number of gaps created from only trying cross-sections along streets.
For hail reports, the swath from edge to edge should be "NO HAIL" to "NO HAIL". For wind reports, a similar technique should be exercised. For flash flood reports, the areas in the affected flooded areas should be explored. There may be a mixture of flood/no flood reports in close proximity, with no pattern except the location's elevation.
Operating over a populated area
When storms are over populated areas, the databases of phone numbers available in Google Earth and Delorme Street Atlas Plus are very useful for finding verification targets. In this situation, the OC's job is primarily to determine the storm of interest and to set up cross sections for examination. They will monitor the progress of the team to complete each cross-section and select a new cross-section. Most of the relevant data concerning call history, hail reports, and current cross-section should be available on the Collaboration Display.
Operating over a rural area
When storms are over rural areas for which county directory information is available, it is more challenging to collect data. The OC or a designated assistant OC will correlate the non-digital or web site sources to a geographic display and enter these phone numbers into the call queue and/or pass them to the data collection team via Google Earth. The results will still be available on the Collaboration Display, but the team will rely on more centralization to determine where to collect data.
Using on-line media sources
Collecting information about wind damage and/or tornado damage is very difficult in the immediate minutes after a storm passes. For this reason, damage locations might be easier determined by monitoring local media sources, such as streaming television and radio coverage, local newspaper sites, and national news and photo wires. Damage data points collected this way should be entered via the standard survey form. Mining of media data sources and follow-up phone calls to fill in gaps in damage swaths may also be conducted the day following a major event.
All data collection entry form entries should be checked for errors immediately before submission. Project participants are enouraged to look through the database during down time to check for errors and make corrections. Adjustments to estimated times are the most frequent source of error, though sometimes location information can be mis-typed. Some "down" days may consist primarily of data quality control.